i915/gem_mmap_gtt: Reduce RSS for fault-concurrent

The intent of exercising parallel page fault is not necessarily to
exercise parallel swap-in (we can safely rely on that being well tested
and is orthogonal to page faulting), but to make sure that our object
and GGTT locking is exercised. We can safely reduce our RSS without loss
of coverage. Furthermore, by using varying sizes we can exercise
different code paths within page faulting, rather than all being
serviced as partial mmaps. Instead of allocating 32 surfaces, each of
16MiB, we allocate 32 surfaces in incremental 512KiB sizes; halving the
memory requirement.

References: https://bugs.freedesktop.org/show_bug.cgi?id=111864
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>

0 files changed, 0 insertions, 0 deletions